Increasing the robustness of software components through automated testing

摘要

Safe software components that are developed, verified, validated and assessed according to DIN EN 50128 [1] and the highest Safety Integrity Level 4 (SIL4) are subject to ever shorter development cycles. In addition, they must take into account increasingly higher quality targets and comply with them, especially with regard to robustness and reliability. The topic of cybersecurity has recently become increasingly important in the safety-related railroad environment. This has been driven by the development of the legal framework both at the European [2] and national levels [3] and in connection with the protection of critical infrastructures. This is particularly relevant due to the fact that (proprietary) bus systems should no longer be used nowadays, but should be replaced with TCP/IP networks where every component can communicate with all the others. While this offers greater flexibility in networking, it also opens up possible attack vectors for unauthorised third-party access. Therefore, manufacturers and operators will not only have to prove the functional safety of critical infrastructures in the future by using selected test methods, but at the same time will have to harden their systems or components in relation to external threats, while also proving that they have done so by using a suitable methodology [4].With regard to this starting point and the goal of achieving complete test coverage more quickly in the future, the need for the further development of existing test methods through the use of new technologies such as machine learning has emerged.